[Thread] MSVC8 - x64 warnings

Hi, I'm compiling boost threads 1.33.1 under msvc8 amd64 targets and there are a bunch of warnings. Simple warnings about loss of data when converting from size_t to unsigned int. The changes below are backward to the normal diff, i.e. the first line is what was added the second is what was removed. I'm sorry that's the way my source control's diff works! tss.hpp: 52c52 < std::size_t m_slot; //This is a "pseudo-slot", not a native slot ---
unsigned int m_slot; //This is a "pseudo-slot", not a native slot
thread.hpp: 73c73 < std::size_t size(); ---
int size();
thread.cpp: 371c371 < std::size_t thread_group::size() ---
int thread_group::size()
recursive_mutex.hpp: 61c61 < std::size_t m_count; ---
unsigned long m_count;
109c109 < std::size_t m_count; ---
unsigned long m_count;
159c159 < std::size_t m_count; ---
unsigned long m_count;
once.cpp (I'm not sure what to do here!): 121,122c121 < // I'm not sure about all this ice_wrapper business, but this seems to work < return InterlockedCompareExchange(dest, exch, cmp); ---
return ice_wrapper(&InterlockedCompareExchange, dest, exch, cmp);
If there is anything else I can do to propogate these, please let me know. Thanks, Sohail

"Sohail Somani" <s.somani@fincad.com> writes:
I've applied a patch to HEAD that calls InterlockedCompareExchange directly if _WIN64 is defined, and uses the existing ice_wrapper code otherwise. This patch has been outstanding for a while. Could someone who's running a 64-bit version of Windows please test this. If it works fine, we could apply it to the release branch. Anthony -- Anthony Williams Software Developer Just Software Solutions Ltd http://www.justsoftwaresolutions.co.uk
participants (2)
-
Anthony Williams
-
Sohail Somani